Hacker News new | ask | show | jobs
by Retric 3554 days ago
Sounds like this just needs automation.

Further people working 100 hour weeks make so many mistakes their work flat out can't be trusted. And should automatically be rejected by any sane review board on that basis alone. Not that it will be but as has been shown many times 'modern' science is fairly broken.
3 comments
dnautics 3554 days ago
when you're at the bleeding edge, if you try to automate your way out of a solution, you will develop poor judgement. One of my grad student cohorts worked on a project where she automated large swaths of her project (which still required about 40-60h prep time). Early on during one of her group meetings, I urged her to take a closer look at her findings and not use automation for her work, but she didn't. I specifically warned her that "she might be chasing ghosts for years", and to not take my warning as any sort of judgment on her character. Later, she found that the thousands of observations she made were artefactual, a result of the automation kit she was using. Nonetheless I have a lot of respect for her for twisting the arm of her PI to allow her to publish the retraction report.

Original report: http://onlinelibrary.wiley.com/doi/10.1002/pro.363/full

Retraction report: http://onlinelibrary.wiley.com/doi/10.1002/pro.2339/full

link
pjlegato 3554 days ago
You are correct that automation would help out a lot technically, at least in some science fields.

There are two cultural issues, though:

1) Automation won't be used in many cases since automation is not how the advisors did things when they got their PhD. They usually don't care if the student has a better idea for how to do things than theirs, even (especially) if the student is technically correct.

2) Even in cases where they do automate some procedure and save a lot of labor, the grad student is simply expected to use that time to do that much more new / additional work.

link
sseagull 3554 days ago
Exactly. They will be put to work figuring out how to automate thing, and we will be back to where we started.

In order to automate something, you have to actually discover and refine the process first

link
jostmey 3554 days ago
Won't happen while labor is cheap
link
pmoriarty 3554 days ago
It will if people start going to jail.
link